Even Nvidia is not immune to the AI chip shortage.
Xinzhou Wu, head of Nvidia’s automotive division, said the company’s automotive division still has to compete internally for access to the GPUs that have made Nvidia the world’s most valuable company.
“Even NVIDIA has a fundamentally limited supply of GPUs for computing,” Wu said on Monday’s episode of The Verge’s “Decoder” podcast.
As demand for Nvidia’s chips continues to soar from AI companies building large data centers, different teams within the company regularly compete for the computing resources needed to train and test their own AI models, Wu said.
“We have internal priorities, and I basically work with my colleagues on an almost weekly basis to determine how we can set aside this additional compute, sometimes for training, sometimes as a testing resource for different work threads within the company,” Wu said.
“And sometimes we need Jensen’s help,” Wu said of the company’s CEO Jensen Huang.
These comments provide a valuable glimpse into how Nvidia is allocating resources within an enterprise where GPU-generated AI is at the root of the boom. Demand for the company’s chips consistently outstrips supply, as companies like OpenAI, Microsoft, Meta, xAI, and Amazon compete to build increasingly large-scale AI models.
Wu said decisions are not driven solely by short-term profits.
When asked how these tradeoffs would be made, he said, “All of the above.” Nvidia is balancing current business needs with long-term strategic opportunities that include entirely new markets that could eventually reach trillions of dollars in value, which Huang called a “zero trillion dollar business,” Wu said.
One of those bets is autonomous driving.
Wu said NVIDIA believes that “everything that moves will be self-driving” and is investing heavily in supplying chips, software, AI models, simulation tools and safety systems for self-driving cars. Although the auto business remains much smaller than Nvidia’s fast-growing data center division, Huang continues to prioritize it.
“We believe strongly in AV, as does Jensen himself. [autonomous vehicle] “We’re basically continuing to invest in this technology and this future, not only through external compute allocations but also fab capacity,” Wu said.
Wu also said that even semiconductor manufacturing capacity has become a new battleground within the company, as demand for Nvidia’s chips continues to soar.